Cavity ringdown measurements of mercury and its hyperfine structures at 254 nm in an atmospheric microwave plasma: spectral interference and analytical performance

Abstract

The plasma-cavity ringdown spectroscopic (Plasma-CRDS) technique has been demonstrated as a powerful tool for elemental and isotopic measurements in recent studies. This work reports the first application of plasma-CRDS to measurements of elemental mercury and its stable isotopes at the 254 nm transition under atmospheric conditions. A microwave-induced plasma (MIP) operating at 80–100 W is used to generate Hg atoms from standard HgCl₂ solutions diluted by 2% nitric acid solvent. It is found that a background absorption, attributed to the overlap of two broadened rovibrational transitions R₂₁(21) and P₁(15) of the OH A-X (3-0) band located at 253.65 nm, generates significant spectral interference with the absorption peak of Hg at 254 nm. With an optimized operating condition, including plasma powers, gas flow rates, and laser beam positions in the plasma, the detection sensitivity of Hg is determined to be 9.1 ng ml⁻¹ in aqueous solution, equivalently 221 pptv in the gas phase; this detection limit is approximately 2-fold higher than the theoretical detection limit, 126 pptv, which was estimated by using the parameters of the instrument system and the calculated absorption cross-section, 2.64 × 10⁻¹⁴ cm² atom⁻¹, of the transition under atmospheric plasma conditions. High-resolution spectral scans show a clear contour of the stable isotopes of the 254 nm transition. The technical challenges encountered and the potential for further development of the Hg analyzer using the MIP-CRDS technique are discussed.

---

Introduction

Mercury is a naturally occurring metallic element which exists in the environment in three forms: elemental mercury, organic and inorganic molecules containing mercury. All three forms of mercury are neurologically toxic. Environmental mercury contamination poses significant threats to human health. Several environmental and occupational health standards have been established for mercury exposure. Mercury concentrations in the atmosphere, from the EPA reference concentration to the safety threshold limit value, are in the range of 0.3 to 25 μg m⁻³, or 3 ppbv to 250 ppbv. The recommended exposure limit for a time-weighted average for an 8 h day is 50 μg m⁻³ or 500 ppbv.¹ There are many research efforts focusing on environmental mercury measurement and control. Most techniques employed for environmental Hg monitoring/analysis are not real-time and are limited by requirements of an off-site sampling and pre-concentration processes. Currently, technologies used or under development for Hg monitoring are cold vapor atomic absorption spectroscopy (CVAAS),² inductively coupled plasma-mass spectrometry (ICP-MS),³ atomic fluorescence spectrometry (AFS),⁴ surface acoustic wave (SAW),⁵ differential optical absorption spectroscopy (DOAS),^6–8 laser atomic absorption spectroscopy (LAAS),⁹ and cavity ringdown spectroscopy (CRDS)-based techniques.^10–15 There are several types of Hg analyzers and/or monitors on the market; and the detection limit of Hg can be as low as 0.01 ng m⁻³.¹⁶ Each instrument/system employing different technologies has its own advantages and disadvantages, depending on the particular applications. However, no single, real-time, portable, sensitive Hg analyzer is available to meet the rising needs in the fields of environmental monitoring, remediation, and management of contaminated sources, such as ground water, testing facilities, and off-gas emissions in coal-fired power plants.

In recent years, we have developed a plasma-cavity ringdown spectroscopy (Plasma-CRDS) technique and demonstrated its application for measurements of several elements and isotopes with detection sensitivities ranging from pg ml⁻¹ to ng ml⁻¹.^17–22 This technique has shown much promise towards the development of a real-time, sensitive, portable analytical tool. In the initial configuration of a plasma-CRDS system, an inductively-coupled plasma (ICP) was coupled with cavity ringdown spectroscopy (ICP-CRDS) and the system demonstrated its technical feasibility by measuring the concentration of lead.¹⁷ The detection sensitivity of the system was later improved by several orders of magnitude through exhaustive effort on the system modification and optimization.¹⁸ Uranium isotopes were also measured by the system at three different transitions; the detection sensitivities were in the 75–150 ng ml⁻¹ levels with the minimum observed uranium isotope shift of 0.28 cm⁻¹ or 3 pm (1 pm = 10⁻¹² m).¹⁹ Driven by the pursuit of low operation costs and portable instrument geometry, a low power, atmospheric microwave-induced plasma (MIP) has recently been introduced as the atomization source. The feasibility of the instrument system (MIP-CRDS) has been demonstrated with measurements of lead at 286 nm.²¹ More recently, continuous wave (cw) diode laser-MIP-CRDS has been explored for strontium (Sr) measurements using a new instrument configuration.²³

Previous research shows that the application of the plasma-CRDS technique to the measurement of each individual element, isotope, or the same element/isotope at different transitions has its own challenges. These challenges arise from differences in the energy level structures of each element, the hyperfine structures of each transition, and the spectral interferences in the different wavelength regions. For instance, the measurement of lead requires the lowest plasma power for a higher ground state population, while a relatively high plasma power is more favorable for the generation of uranium atoms/ions.^18,19 Strong background spectral interference hinders the improvement of detection sensitivity for ²³⁸U isotope, while it has no effect on the ²³⁵U isotope.¹⁹

This work reports the first measurement of elemental mercury and its stable isotopes at the 254 nm transition using MIP-CRDS. The initial detection sensitivities of the elemental Hg measurements and the observation of stable isotopic structures at 254 nm are reported. Discussion of the influence of the spectral interferences, the spectral linewidth broadening, and the effect of sampling handling on the system detection sensitivity are presented. These results will be a useful database in the areas of plasma design and sample introduction for further studies of Hg using the MIP-CRDS technique towards the development of a real-time MIP-CRDS analyzer.

Experimental

The entire plasma-CRDS system consists of four major sections: optical system, plasma source, sample introduction, and data acquisition portion, which are briefly described below. A detailed description of the MIP-CRDS system can be found in previous publications.^21,23

Optical system

The optical system includes a pulsed laser source, which provides UV radiation around the 254 nm spectral region. The UV beam was generated through the frequency doubled output of a dye laser (NarrowScan) pumped by a Nd:YAG laser (Continuum 8020, at 355 nm) at a repetition rate of 20 Hz. The laser dye was coumarin 500 (Exciton). The laser linewidth in the UV, extrapolated from the linewidth (<0.05 cm⁻¹) at 590 nm, is approximately 0.08 cm⁻¹ (∼0.0017 nm at 254 nm). The minimum scanning step of the dye laser is 0.0006 nm, equivalently 0.0003 nm in the UV. This narrow linewidth laser source assures the capability of obtaining a high-resolution spectrum of Hg. The UV radiation was imaged by a telescope into a ringdown cavity, which was formed by a pair of high reflectivity mirrors separated by 78 cm. The mirrors have a plano-concave configuration with a radius of curvature of 6 meters and a quoted reflectivity of >99.75% at 254 nm, but the actual mirror reflectivity at 254 nm was measured to be 99.67% due to the degradation of the coating after extensive use. The optical signal was detected by a photomultiplier tube (PMT-Hamamatsu R928), and a 10 nm bandpass interference filter (CVI Laser) was mounted in the front of the PMT window to reject emissions from the plasma. A monochrometer (ARC SpectraPro-500) coupled with a charged-coupled device (CCD) array (Princeton Instruments TEA/CCD-1024) was used to monitor plasma performance by recording emission spectra of Hg and the plasma background.

Plasma source and operating conditions

A candle-shaped microwave induced plasma torch, connected to a 2450 MHz microwave power supply, was located in the middle of the cavity (between the two mirror surfaces). The vertical, h, and lateral, x, position of the plasma torch could be tuned by a home-made x–y mount with a spatial accuracy of ± 0.5 mm. The typical plasma operating conditions for the Hg measurements are listed in Table 1. The height of the laser beam in the plasma is defined as the distance between the laser beam and the top surface of the plasma torch. The single path-length of the laser beam through the plasma is typically 1–8 mm, depending on the height of the laser beam. The laser was positioned to pass directly over the center of the sample injection tubing oracle, which was centered in the plasma torch. Argon (Air Gas, 99.99%) was used as the plasma supporting gas.

Table 1 MIP operation conditions

Plasma
Microwave plasma power	80 W
Plasma supporting gas flow rate	0.35 l min^−1
Plasma central gas flow rate	0.50 l min^−1
Sampling
Sample up-take rate	1.0 ml min^−1
Heating temperature of the ultrasonic nebulizer chamber	120 °C
Cooling temperature of the ultrasonic nebulizer desolvator	−5 °C
Heating temperature of the membrane device	80 °C
Ar gas flow rate in the drier	0.5 l min^−1

---

Standards and reagents

Mercury sample solutions at multiple concentrations were prepared by diluting the standard solution (1000 μg ml⁻¹, Absolute Standard Inc, Hamden, CT) with a 2% nitric acid solution. The sample solutions were pumped to an ultrasonic nebulizer (CETAC U-5000AT⁺) to generate an aerosol, which was, in turn, carried by argon gas through a moisture drier before being injected into the plasma torch. The nebulizer has a typical gas phase conversion efficiency of 10%. Throughout the experiment, a ventilation system was operating to vent the plasma exhaust to a designated avenue. The nebulizer and sample bottles were housed in a plexi-glass cabinet, which was also continuously vented by the same system. A 100 ng ml⁻¹ gold solution was prepared to flush the sampling tubing (110 cm in length) to remove the Hg residue in the tubing after each experiment. A certified, environmental professional regularly disposed of the solid waste and residual Hg solutions.

Data acquisition

The ringdown signal from the PMT was monitored by an oscilloscope (Tektronix, TDS 410A) that was interfaced to a computer for data processing. Each data point, a single ringdown time, was typically generated by averaging 50–100 ringdown waveforms followed by the ringdown fitting routine.^18,21,23

Results and discussions

Broadband spectral scan and location of Hg 254 nm transition

Fig. 1 shows the cavity ringdown spectrum obtained in the vicinity of 254 nm under ambient conditions without the plasma operating. The purpose of the broadband scan is to investigate all possible spectral interferences from the atmosphere. The measured spectrum is assigned to Herzberg I bands (7–0) and (8–0) of the O₂ A³ Σ⁺_u–X Σ⁻_g system with the help of the high-resolution rotational analysis.^24–26 No spectral absorption attributed to other atmospheric molecules was found. The Hg 254 nm transition is located between two rotational branches with rotational quantum numbers, N″ = 19 and 21. The pattern of the spectral intensity distribution and reproducibility of the spectrum in Fig. 1 demonstrates the high quality performance of the ringdown system in terms of the signal-to-noise ratio of the spectrum; e.g., the weak transitions between peaks at N″ = 13, 15 and N″ = 7, 9 are repeatedly observed. The spectrum in Fig. 1 was not corrected for the wavelength dependence of the mirror reflectivity since the covered spectral region was only ∼2 nm.

Fig. 1 Cavity ringdown spectrum measured in the vicinity of 254 nm under ambient conditions without the plasma operating. The spectrum is attributed to Herzberg I bands (7–0) and (8–0) of the O₂ A ³Σ⁺_u–X Σ⁻_g system. The arrow indicates the location of the Hg 254 nm transition.

Background spectral interference

Fig. 2 shows the high-resolution spectra obtained in the range of 253.56–253.74 nm where the Hg transition (253.652 nm) is located.²⁷ The spectra were obtained with the minimum scanning step of the dye laser, 0.0006 nm (equivalently, 0.0003 nm in the UV). Fig. 2(a) shows that there is no spectral inference at 253.652 nm when the plasma is off. The left-hand part of the spectrum in Fig. 2(a) is attributed to Q multiplet transitions of the O₂ A ← X (7–0) band at N″ = 19, as they are assigned in the figure. The spectrum in Fig. 2(b) was obtained when the plasma was on but no sample was injected to the plasma. In Fig. 2(b), except for the O₂ transitions in the shorter wavelength side of the spectrum, there are four absorption peaks located at approximately 253.65 nm. In order to determine the absorption sources of these four peaks, the following experimental procedures in spectral scans were performed: (1) the plasma was on with only the argon supporting gas flowing, and (2) the plasma was on with the argon supporting gas and the injection of blank solution (2% nitric acid in DI-water). In both cases, the four absorption peaks were observed and reproducible.

Fig. 2 High resolution cavity ringdown spectra around the Hg 254 nm transition. (a) The plasma was off; (b) the plasma was on. Peak 3 in Fig. 2(b) is attributed to the overlap of two broadened rovibrational transitions, R₂₁(21) and P₁(15) of the OH A-X (3-0) band located at 253.65 nm.

Of the four peaks, Peak 3 is located at 253.65 nm, approximately the same position of the broadened Hg 254 nm transition. The linewidth (full width at half maximum, FWHM) of Peak 3 is ∼0.02 nm, which is unlikely from an atomic transition. A typical linewidth of an atomic transition without the hyperfine structures under plasma conditions is ∼0.0025 nm, and ∼0.003–0.005 nm with the hyperfine structures. Based on the spectral simulation using LIFBASE (version 2.0),²⁸ Peak 3 is assigned to the rovibrational transitions R₂₁(21) and P₁(15) of the OH A-X (3-0) band. Under the plasma conditions, these two broadened transitions are overlapped into a single peak and located at 253.65 nm. The Hg transition, 253.652 nm, is located on the shoulder of this interference peak. These two peaks completely overlap when the Hg transition is subject to the linewidth broadening effect under the plasma conditions. This interference does not affect the measurement of Hg when the sample concentrations are high, e.g., in the hundreds of ng ml⁻¹ levels, since a good signal-to-noise ratio can be still obtained after the subtraction of the interference contribution from the strong ringdown signal. However, the interference does influence the determination of the detection limit of the system, since the comparatively poorer signal-to-noise ratio is obtained after the subtraction when the sample concentrations are close to the detection limit, e.g., below tens of ng ml⁻¹ levels, in which the weak ringdown signals are observed. A similar interference effect was observed in the measurement of the ²³⁸U isotope using ICP-CRDS,¹⁹ in which the detection limit was affected by a factor of 2.

Peak 4 has a similar linewidth to Peak 3 and is also not likely to be from an atomic transition. Spectral calculations of several other radicals that are very likely generated in atmospheric plasmas, such as CH and CN, were also performed. However, no reliable assignment can be attributed to Peak 4. Peaks 1 and 2 have a narrower linewidth, thus possibly from atomic transitions. However, efforts to assign these two peaks to elemental transitions from elements such as argon, copper, lead, and uranium, failed. These elements are most likely generated from the plasma supporting gas, the plasma torch (Cu), and the presence of residual lead and uranium from previous experiments. Additionally, oxygen mixed with argon was injected into the plasma to investigate the possibility that these peaks (Peaks 1, 2, and 4) may originate from plasma-induced radicals or oxygen atoms. No difference was observed in the background spectral scans obtained with and without O₂ injections. Peaks 1, 2, and 4 in Fig. 2(b) remain unassigned in this work.

Measurements of Hg at 254 nm

Mercury sample solutions with different concentrations were prepared through dilutions of a standard stock solution, as described in the experimental section. Fig. 3 shows the high resolution spectral scans obtained under three different operation conditions: plasma off, plasma on with a blank solution, and plasma on with a 240 ng ml⁻¹ Hg sample solution. Detection of Hg at 254 nm is complicated by the OH transition, which induces significant background interference at this wavelength, as is clearly depicted in Fig. 3. In order to obtain an actual profile of the Hg transition, the background absorption must be subtracted.

Fig. 3 Spectral scan around 254 nm. The top represents the spectral scan when the plasma was off; the middle denotes the scan when the plasma was on with a blank solution of 2% nitric acid in DI water; the bottom denotes the scan when the plasma was on with the Hg sample solution of 240 ng ml⁻¹.

For a given ringdown system (a fixed cavity length and mirror reflectivity), the detection sensitivity is experimentally reliant upon the ringdown baseline noise, which is defined as σ/τ, where σ and τ are the standard deviation of the ringdown time and the ringdown time, respectively. In a plasma-CRDS system, the lowest baseline noise can be achieved through a comprehensive system optimization which includes optimization of the plasma power, gas flow rates, and laser beam position (h, x) in the plasma.^18,21 For different elements of interest, each of these parameters has to be optimized individually. In the experiment, the overall performance of the optimized MIP-CRDS system is evaluated by the lowest ringdown baseline noise and the strongest ringdown signal (the largest ringdown time difference, Δτ). The parameters listed in Table 1 represent the optimized operating conditions for Hg measurements using this plasma-CRDS system. As an example to show the influence of the optimization on the system performance, Fig. 4 depicts the height dependence of the Hg ringdown signal. The different heights of the laser beam in the plasma can affect the signal intensity by a factor of 6. In Hg measurements, the laser beam was positioned through the center of the plasma, 6 mm above the plasma torch. With the optimized system, the averaged ringdown time at 254 nm when the plasma was off was 780 ns and the standard deviation was 2.34 ns; this result yielded a baseline noise of 0.3%. When the plasma was on, the averaged ringdown time at 254 nm dropped to 450 ns due to the plasma-induced scattering loss and the presence of the OH absorption; the standard deviation was 2.25 ns, which corresponds to a baseline noise of 0.5%.

Fig. 4 The observed ringdown signal is dependent upon on the laser beam position (height, h) in the plasma. The laser beam was aligned to pass through the center of the plasma. The sample concentration was 240 ng ml⁻¹.

Determination of the system detection limit is typically achieved by obtaining a calibration curve. However, in this experiment, obtaining a quality calibration curve was hindered by the OH background interference and additional adsorption interference of the Hg sample on the wall of the sample introduction tubing. These two factors heavily influenced the system performance. When the sample concentrations were low, e.g., below 50 ng ml⁻¹, the weak ringdown signal (the ringdown time difference, Δτ) yielded a worse signal-to-noise ratio; and the measurement uncertainty generated by the subtraction of the OH interference was up to 20%. This large error is due to the fact that the laser wavelength was tuned to 253.652 nm, right on the shoulder of the OH interference, Peak 3 in Fig. 2(b). A 0.002 nm drift of the laser wavelength can introduce a measurement error up to 20% in the spectral subtraction when the sample concentrations are low. In the experiment, the laser wavelength was simultaneously monitored by a UV wavemeter (Burleigh WA-5500) with an accuracy of ±0.001 nm. In order to insure that the Hg adsorbed to the sample tubing would not contaminate subsequent samples, the 110 cm long sample introduction tubing was flushed with a 100 ng ml⁻¹ gold solution prior to and following each sample. Note that using a gold solution to clean the mercury sample residuals in the sampling tube is found to be a very effective way, which is a routinely employed in the elemental analysis using an ICP-MS in our analytical laboratory. The measurement procedure followed a cycle: gold solution, blank, sample solution, and gold solution. The efficiency of the flush was evaluated by monitoring ringdown signals and the emission spectra obtained by the monochromator. The effect of the flushing process was relevant to the concentration of the gold solution used, the flushing time, and the Hg sample solution concentration used. When the sample solution concentrations were low, e.g., 5–50 ng ml⁻¹, the cleaning process could yield an under estimation of the detection sensitivity as the Hg amalgamated to the residual gold. For instance, when the residual gold solution was present, then the ringdown signal of a sample solution of 5 ng ml⁻¹ was attenuated or not even observed. The measurement uncertainty was as large as 20% in the lower concentrations. In the experiment, no reliable calibration curve was obtained when the sample concentrations were below 50 ng ml⁻¹. However, a linear response between the ringdown times and the concentrations of the sample solutions at 60, 120, and 240 ng ml⁻¹ was observed.

In order to estimate the experimental detection limit, ringdown measurements were repeatedly performed with the 240 ng ml⁻¹ sample solution. The difference in the averaged ringdown times with and without the sample solution is 175 ± 3 ns. Based on the standard 3σ criteria, the minimum detectable ringdown time difference is 6.75 ns, which corresponds to a 9.1 ng ml⁻¹ detection limit, or 221 pptv in the gas phase, extrapolated from the ringdown time difference, 175 ns, when the sample solution was 240 ng ml⁻¹. This estimated number is based on the assumption of a linear response to the concentration in the range of 0–240 ng ml⁻¹. In these experiments, absorption saturation became apparent when the sample concentrations were higher than 1000 ng ml⁻¹.

Theoretical detection limit

Detection limits of concentrations of sample solutions are quantitatively different from detection limits of Hg concentrations in the gas phase due to the fact that there is a conversion efficiency of the concentration of the sample solution to the concentration in gas phase. The measured ringdown signal is a direct reflection of the concentration of elemental Hg in the gas phase in the plasma, rather than the Hg concentrations in the sample solutions. The conversion of the detection limit in the sample solution to the detection limit in the gas phase depends on the conversion efficiency of the ultrasonic nebulizer, the pumping rate of the sample solution, and the carrier gas flow rate.^18,24 Using the conversion equation tabulated in Table 2, a conversion relation for Hg is determined: 1 ng ml⁻¹ in aqueous solution = 25.8 pptv in the gas.

Table 2 Parameters for converting concentrations from aqueous solution to gas phase

Definition	Parameter	Used in the work	Units
Injection rate of solution sample	m	1.0	ml min^−1
Sample concentration	c s	To be determined	g ml^−1
Sample density	n	Measured by CRDS	atom cm^−3
Carrying gas flow rate	V	0.5	l min^−1
Nebulizer conversion efficiency	η	10	%
Molar mass	M	Hg, 200.59	g
Avogadro constant	N	6.02 × 10^23	mol^−1
Laser beam path-length in the plasma	l	6	mm
Calculated absorption cross-section	σ(ν)		cm^2
Minimum detectable absorbance	A= (1 − R)3στ/τ	4.95 × 10^−5	au
Mirror reflectivity	R	99.67 (R (new) = 99.75)	%
Baseline noise	σ τ /τ	0.5	%
Conversion formula		Ref. 23

---

The theoretical detection limit for Hg in the gas phase under the tabulated plasma conditions can be estimated using the absorption cross-section of the Hg 254 nm transition, the measured ringdown baseline noise, and the system parameters, such as mirror reflectivity and the effective laser beam path-length in the plasma. The absorption cross-section of the Hg 254 nm transition measured in ambient conditions is 3.3 × 10⁻¹⁴ cm² atom⁻¹.²⁹ However, this number cannot be directly used for the estimation of the theoretical detection limit in this high temperature measurement since the Doppler broadening effect has to be taken into account in determining the linewidth broadening effect as well as the lineshape, which directly affect the absorption cross-section.³⁰ In addition to the broadening effect, the Hg transition at 254 nm has neither a pure Lorentzian nor a pure Gaussian lineshape due to the isotopic structures. The measured linewidth (FWHM) under the current plasma conditions with an estimated gas kinetic temperature of 2800 K,^22,31 is 5.0 pm (0.78 cm⁻¹), which is approximately three times larger than the one obtained in ambient conditions, 1.6 pm (0.25 cm⁻¹).¹³ A typical plasma broadened linewidth for an atomic transition in the UV spectral region is ∼2.5 pm (∼0.4 cm⁻¹). This large linewidth of the Hg transition is due to the isotopic structures. PeakFit,^18,32 was used to fit the measured lineshape of the transition obtained with the 240 ng ml⁻¹ sample solution to a Voigt profile and yielded the fitted result with a coefficient determination R² = 0.98. The fitted linewidth components from Gaussian and Lorentzian broadening are 2.95 pm and 1.74 pm, respectively. These results yielded a value of the Voigt function, V(a, 0), to be 0.63, where a is defined as , and w_L and w_G represent the Lorentzian and Doppler components, respectively. Based on these results, the calculated absorption cross-section is 2.64 × 10⁻¹⁴ cm² atom⁻¹, which is slightly smaller than the reported value obtained in ambient conditions, 3.3 × 10⁻¹⁴ cm² atom⁻¹.¹³

Using the calculated absorption cross-section, 2.64 × 10⁻¹⁴ cm² atom⁻¹, the measured mirror reflectivity, 99.67%, the effective single path-length of the laser beam through the plasma, 6 mm, and the ringdown baseline noise, 0.5%, the theoretical detection limit of the instrument system for Hg, based on the 3σ criteria, is determined to be 5.2 ng ml⁻¹, or 126 pptv. This estimated detection limit is 2-fold lower than the experimental one.

Isotopic structures

The capability of mercury isotopic measurements is a significant merit in real applications such as Hg monitoring and control, in which a seeded Hg isotope may be used as a tracer to track the migration and fate of the environmental Hg in contaminated sources. Mercury has seven stable isotopes.^33–35 Five spectral peaks attributed to the isotopes are typically observed at 254 nm. Under atmospheric conditions, the transitions of these isotopes are partially overlapped; and at high temperatures, Doppler broadening makes the isotopic transitions even less resolved. Fig. 5 shows a high-resolution spectrum of Hg at 254 nm. The contour of the isotopic structures is apparent while the transitions of each individual isotope are not resolved. The five peaks can be identified and assigned to specific isotopes or combinations of them based on the well-documented isotopic abundance ratios and transition positions. Except for Peaks 3 and 4, each individual Peak, 1, 2, and 5, is composed of transitions from more than one isotope. The contour of the transition at 254 nm shows the isotopic structure more clearly than that reported in previous ringdown measurements where Hg atoms were generated by a chemistry method.¹³ The contour of the isotopic structures observed in Fig. 5 is very similar to the one observed using atmospheric argon ICP-CRDS.²⁰ The abundance ratios of the five isotopic groups are estimated using the relative signal intensity ratio, and the results are tabulated in Table 3. The ratios are in good agreement with the theoretical values, except for Peak 1, which has a large measurement uncertainty. This is partially due to the lower abundance of this peak and the additional error generated by background subtraction. To the longer wavelength side of Peak 5 in Fig. 5, there is a small but noticeable shoulder as marked by an *. This small hump is not attributed to spectral noise generated through the spectral background subtraction since it is repeatable in the experiment. A spectral assignment of this peak needs to be addressed in future work.

Fig. 5 Hyperfine structures of the Hg 254 nm transition measured using the atmospheric argon MIP-CRDS. The laser linewidth was 0.08 cm⁻¹ (0.0017 nm). The scanning step of the dye laser was 0.0006 nm, which corresponds to 0.0003 nm in the UV.

Table 3 Measured abundance ratio and the isotopic assignment

Line position	Wavelength/nm (air)	Exp. ratio (%)	Theo. ratio (%)	Isotope assignments
1	253.647	6.9	13.4	201(c) + 199(a)
2	253.650	12.6	14.4	201(b) + 198
3	253.652	28.0	23.1	200
4	253.653	32.1	29.8	202
5	253.654	20.3	19.1	204 + 201(a) + 199(b)
*	253.656			Unknown

---

Technical challenges and improvement potential

In the efforts to develop an MIP-CRDS analyzer for atomic Hg measurements, the technical challenges exposed in this work need to be further addressed in subsequent studies. The first one is the elimination of the OH interference; the second one is the minimization of the Hg adsorptive effect in the sample introduction tubing. These two factors influence both the improvement of the detection sensitivity as well as the ability to accurately determine the Hg concentration. In an atmospheric plasma (a candle-shaped ICP or MIP), OH is generated partially by the dissociation of the moisture in the air surrounding the plasma.²² Due to the variation of the moisture content in the atmosphere, the observed signal intensity is affected even though the other measurement and operating conditions of the system are the same. As discussed earlier, a large measurement error can be introduced through the background subtraction when the sample concentrations are low. Therefore, in order to improve the system performance, the OH interference has to be eliminated. Our very recent study shows that the OH interference is completely eliminated by using a low temperature, long tube-shaped MIP source.³⁶ The adsorption of Hg in the sample introduction tubing can be minimized by using a shorter section of tubing, which would require the instrument configuration to be rearranged. The third challenge in the effort to develop a portable analyzer using MIP-CRDS is the lack of a portable and low cost UV laser source operating at 254 nm. A portable, narrow linewidth laser source operating at 254 nm can be generated through frequency doubling of a diode pumped laser; such a laser source is already on the market; however, it is still rather expensive.

Additionally, isotopic measurements of Hg under atmospheric pressure and high plasma temperatures cannot provide an accurate determination of the isotopic abundance due to the overlap of the isotopic structures resulting from the pressure broadening (Lorentzian broadening) and the temperature broadening (Doppler broadening) under the atmospheric plasma conditions. One possibility to circumvent the broadening effects is to introduce a low pressure chamber into the MIP-CRDS system, thus the isotopic abundance can be determined from the well-resolved isotopic peaks. Additionally, using a narrow linewidth diode laser will also help resolve the isotopic structures. The laser linewidth at 254 nm in the current system is 0.08 cm⁻¹, or 0.5 pm. A frequency doubled diode laser provides a UV beam with a linewidth of <0.001 cm⁻¹, or 0.006 pm.

Finally, the improvement potential of the detection sensitivity can be further explored with the consideration of following three factors: an increased path-length of the laser beam in the plasma, higher reflectivity ringdown mirrors (>99.67%), and elimination of the spectral interference from OH radicals. The ultimate detection sensitivity of an Hg analyzer using the MIP-CRDS technique is expected to be comparable to the detection limits obtained with a typical ICP-MS system, 0.1–10 pptv.

Conclusions

The first effort of applying the MIP-CRD technique to the measurement of atomic Hg and its stable isotopes is reported. The detection sensitivity of Hg is estimated to be 9.1 ng ml⁻¹ in solution, or 221 pptv in the gas phase, using the current MIP-CRDS system with a mirror reflectivity of 99.67% and an effective laser beam path-length of 6 mm in the plasma. The experimental detection limit is approximately 2-fold higher than the theoretical one that is estimated by using the instrument parameters and the calculated absorption cross-section, 2.64 × 10⁻¹⁴ cm² atom⁻¹, of the Hg transition at 254 nm under the specified plasma conditions. The spectral interference from the rovibrational transitions of OH radicals generated in the MIP is reported. This background spectral interference, together with the sample contamination in the sample introduction tubing, affects the detection sensitivity and measurement accuracy of the MIP-CRDS system using a candle-shaped MIP source. This work also demonstrates the potential capability of measuring the isotopic abundance of Hg using the MIP-CRDS technique. The results obtained in this work constitute a valuable foundation to help address the technical challenges and assist in the future development of a mercury analyzer using MIP-CRDS. Efforts to improve the system performance by using different types of MIP sources and different instrument configurations are underway.

Acknowledgements

The authors are thankful for the support from the US Department Energy, Office of Science, under grant DE-FG07-02ER63515 at Mississippi State University and grant 86680 at Los Alamos National Laboratory.

References

1. US EPA, Integrated Information System, Mercury, elemental (CASRN 7439-97-6).
2. For example B. Welz and M. Schuber-Jacobs, Fresenius’ Z. Anal. Chem., 1988, 33, 324.
3. R. Brown, D. J. Gray and D. Tye, Water Air Soil Pollut., 1995, 80, 1237.
4. H. Tanaka, M. Kouno, H. Morita, K. Okamoto and S. Shimomura, Anal. Sci., 1992, 8, 1047.
5. Sensor R&D Corp. Maine, www.srdcorp.com.
6. H. Edner, A. Sunesson, S. Svanberg, L. Unéus and S. Wallin, Appl. Opt., 1986, 25, 403.
7. U. Platt, D. Perner and H. W. Patz, J. Geophys. Res., 1979, 84, 6329.
8. U. Platt, D. Perner, A. M. Winer, G. W. Harris and J. N. Pitts, Nature, 1980, 285, 312.
9. W. Wang, R. H. Hammond, M. M. Feijer, C. H. Ahn, M. R. Beasley, M. N. Levenson and M. L. Bortz, Appl. Phys. Lett., 1995, 67, 1375.
10. A. O’Keefe and D. A. G. Deacon, Rev. Sci. Instrum., 1988, 59, 2544.
11. ACS symposium Series 720: Cavity-Ringdown Spectroscopy: An Ultratrace-Absorption Measurement Technique, ed. K. W. Busch and M. A. Busch, Oxford University Press, 1999.
12. R. T. Jongma, M. G. H. Boogaarts, I. Holleman and G. Meijer, Rev. Sci. Instrum., 1995, 66, 2821.
13. S. Spuler, M. Linne, A. Sappey and S. Snyder, Appl. Opt., 2000, 39, 2480.
14. S. Tao, F. Mazzotti, C. B. Winstead and G. P. Miller, Analyst, 2000, 125, 1021.
15. G. P. Miller, Sensor Research & Development Inc. Maine, USA, 2003, personal communication.
16. Mercury Analyzer for Gas Mercury/WA-4, Brandt Instrument, http://www.brandtinst.com/nic/wa-4.
17. G. P. Miller and C. B. Winstead, J. Anal. At. Spectrom., 1997, 12, 907.
18. Chuji Wang, F. J. Mazzotti, G. P. Miller and C. B. Winstead, Appl. Spectrosc., 2002, 56, 386.
19. Chuji Wang, F. J. Mazzotti, G. P. Miller and C. B. Winstead, Appl. Spectrosc., 2003, 57, 1167.
20. Chuji Wang, F. J. Mazzotti, J. Mierzwa, G. P. Miller, Yixiang Duan and C. B. Winstead, Pittcon 2002, New Orleans, LA.
21. Yixiang Duan, Chuji Wang and C. B. Winstead, Anal. Chem., 2003, 75, 2105.
22. Chuji Wang, F. J. Mazzotti, S. P. Koirala, C. B. Winstead and G. P. Miller, Appl. Spectrosc., 2004, 58, 741.
23. Chuji Wang, S. P. Koirala, S. T. Scherrer, Y. Duan and C. B. Winstead, Rev. Sci. Instrum., 2004, 75, 1305.
24. K. Yoshino, J. E. Murray, J. R. Esmond, Y. Sun, W. H. Parkinson, A. P. Thorne, R. C. M. Learner and G. Cox, Can. J. Phys., 1994, 72, 1101.
25. K. Yoshino, J. R. Esmond, J. E. Murray, W. H. Parkinson, A. P. Thorne, R. C. M. Learner and G. Cox, J. Chem. Phys., 1995, 103, 1243.
26. D. L. Huestis, R. A. Copeland, K. Knutsen, T. G. Slanger, R. T. Jongma, M. G. H. Boogaarts and G. Meijer, Can. J. Phys., 1994, 72, 1109.
27. M.I.T. Wavelength Tables, Cambrige, Massachusetts and London, UK, 1969.
28. J. Luque, LIFBASE, Window Version 2.0.130, 2003.
29. H. Edner, G. Faris, A. Sunesson and S. Svanberg, Appl. Opt., 1989, 28, 921.
30. A. Montaser and D. W. Golightly, Inductively Coupled Plasma in Analytical Atomic Spectrometry, VCH Publishers, Inc., 1992.
31. Y. Duan, S. T. Scherrer, S. P. Koirala, Chuji Wang and C. B. Winstead, Anal. Chim. Acta, 2005, 532, 47.
32. PeakFitTM 4.0 for Windows User’s Manual, Copyright 1997 by SPSS Inc .
33. V. Kaufman, J. Opt. Soc. Am., 1962, 52, 866.
34. W. G. Schreiber Jr., J. Opt. Soc. Am., 1963, 53, 1055.
35. Y. Nishimura and T. Fujimoto, Appl. Phys. B, 1985, 38, 91.
36. Y. Duan, Chuji Wang, S. T. Scherrer and C. B. Winstead, “Development of Alternative Plasma Sources for Mercury Cavity Ring-Down Measurement”, Anal. Chem., in the press.

---